Newsgroups: comp.ai.genetic
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!hookup!swrinde!howland.reston.ans.net!Germany.EU.net!EU.net!sun4nl!news.nic.surfnet.nl!utciva.civ.utwente.nl!usenet
From: "Albert van Breemen" <s9101799@mail.student.utwente.nl>
Subject: Re: About Hollands book 'Adaptation in ...'
To: kslee@srgcentre
Message-ID: <53350.s9101799@mail.student.utwente.nl>
X-Minuet-Version: Minuet1.0_Beta_17
Sender: usenet@Utwente.NL ((UT-utciva)USENET News System)
X-Popmail-Charset: English
Reply-To: <a.j.n.vanbreemen@student.utwente.nl>
Organization: University of Twente, Enschede, The Netherlands
Date: Mon, 5 Dec 1994 17:11:26 GMT
Lines: 51

On Sun, 4 Dec 1994 11:55:17 GMT, 
Lee Kwok Shing ?***? ?~?  <kslee@srgcentre> wrote:

>"Albert van Breemen" (s9101799@mail.student.utwente.nl) wrote:
>:Q> Introducing me: Albert van Breemen, student with the Department of Electrical 
>:Q>                 Engineerinf, University of Twente, the Netherlands
>:Q> Interest:       AI, at the moment specially GA
>
>:Q> Problem:
>
>:Q> I've read the most of Hollands book 'Adaptation in natural and 
>:Q> artificial systems'. While reading the book the following question 
>:Q> did arise: GA works only good if their is some correlation between f(c) and
>:Q> c, with c a structure of A. If their is no correlation then the search is
>:Q> just random (This is just mine opinion which I have not proved or neither 
>:Q> I've read any prove, but I want this to prove mathematicaly soon). If their 
>:Q> is some correlation then I mean:
>        
>        I think your observation is correct. To think in another way,
>if f(c) is not related to c, it's just an random function. Any other methods
>can tackle the problem ? (I'm still thinking about it, but I believe no
>method can solve them as well.)
>        Please comment !

Well I'm not a hero at probabily calculation, and I think that is the way 
the prove that if there is now correlation then the rearch is random! Well 
there is a way, I think, to solve this problem. Maybe it's simple, but if 
you could find a way to improve the correlation between c and f(c) then 
you'll get a bias for the search! In Holland's book [Hol75] he describe a 
'broadcast language'. It's like enzymes in a cell: the 'broadcast 
language'-enzyme looks at the chromosome and trie to find some pattern in 
it. If the chromosome holds this pattern then it modify's the structure of 
the chromosome. In this way it tries to improve the correlation. This is 
what I made up of the book but I don't know if this is what Holland meant, 
but it is a good idee! There is here one probleem: How to choose your 
"broadcast language" (b.c.)? This b.c. are also string, so Hollands 
proposal was to optimize this with GA during the evolution of your problem. 
But what if there is now correlation between the b.c. and the performance! 
So I doub that is b.c. will work. Is there someone who has used this 
"broadcast language" and what were the results!.



================================================
A.J.N. van Breemen
lipperkerkstr 128
7511 DD  ENSCHEDE, The Netherlands
tel: 053-317819
email: a.j.n.vanbreemen@student.utwente.nl
================================================
       
